A national survey is mainly designed to produce reliable estimates of target variables of the population at national and regional levels. However, at the unplanned domains, lower governmental administration layers estimate are unreliable as the sample sizes are small, which leads to large sampling errors. Ethiopian administrative zones are unplanned domains that produce unreliable estimates. We applied small area estimations to overcome the problem of having such unreliable estimates by linking the survey data to the census data. This dissertation aims to improve the precision of undernutrition estimates at zonal level using small area estimation (SAE) based on the Ethiopian demographic and health survey (DHS) data. This dissertation applied several statistical models to further enhance the direct survey estimates. The models employed include multivariate SAE, spatial SAE, spatial nonstationarity SAE, Spatio temporal SAE, and Bayesian spatial SAE. The statistical model assumptions of these models are checked and met. The coefficient of variation (CV) and the root mean square errors (MSE) measured the improvements in undernutrition estimates in Ethiopian zones. According to the results, model-based small area estimates have better precision than direct survey estimates. Therefore, the findings of model-based small area estimates are more precise and reliable than the direct survey estimates. Finally, the findings revealed that more reliable and precise disaggregated undernutrition statistics are produced at small area levels (zones). These estimates will save the government from conducting surveys at the zonal level with abundant resources. The results also provide useful information to the government's planners, policymakers, and legislative organs for effective policy formulation and budget allocation in Ethiopian zones. It is recommended that further studies using more advanced statistical models and different socio economic problems can be conducted under the unit level small area estimatio